CONF Calinon_HFR_2016/IDIAP Stochastic learning and control in multiple coordinate systems Calinon, Sylvain Intl Workshop on Human-Friendly Robotics Genoa, Italy 2016 1-5 A probabilistic interpretation of model predictive control is presented, enabling extensions to multiple coordinate systems. The resulting controller follows a minimal intervention principle, by learning and retrieving movements through the coordination of several frames of reference. When combined with a generative model, the approach can be used in various human-robot applications that are discussed in the paper.